联系我们

论文查询结果	返回搜索

论文编号：	15994
作者编号：	2120243877
上传时间：	2026/6/2 12:26:40
中文题目：	大学生人智交互对心理压力的缓解效应及作用机制研究
英文题目：	Research on the Alleviation Effect and Mechanism of Psychological Stress through Human-AI Interaction among College Students
指导老师：	张丹
中文关键字：	人智交互；心理压力；信息行为；生成式 AI；压力缓解
英文关键字：	Human-AI interaction; Psychological stress; Information behavior; Generative AI; Stress relief
中文摘要：	随着生成式人工智能的快速发展和大学生心理健康问题的日益突出，人智交互在心理健康领域展现出巨大潜力。已有研究主要关注了用户对AI心理健康服务的接受度与态度，但人智交互缓解心理压力的过程机制尚未得到系统解构，用户需求与AI供给的匹配关系、用户主观体验、交互对压力缓解的作用机制等问题尚不清晰。本研究聚焦于压力状态下大学生与生成式AI的交互行为，综合运用实验法、半结构深度访谈法和内容分析法，对25位用户的深度访谈资料进行系统分析，对25组用户与DeepSeek的对话文本进行内容分析，辅以压力前后测的量化评估。研究从压力缓解效果分析、主观体验建构、客观行为轨迹三个层次依次展开。 研究发现，在压力缓解方面，人智交互对压力缓解具有效果；人智交互通过认知重构、情绪支持、行动赋能、支持感知、决策清晰化、正常化六种机制实现压力缓解，不同压力类型呈现差异化的路径选择。主观体验层面，用户体验由信息需求预设、信息交互行为、信息呈现感知、信息效用评估、信息源可信度评估五个核心概念构成体验过程，并呈现信息处理与关系建构双路径结构。信息处理路径遵循供需匹配逻辑指向问题解决，关系建构路径遵循信任累积逻辑指向可信度评估，两条路径并行运作、相互影响。此外，用户初次进入交互使用AI缓解压力受推力、拉力、阻力三类力量共同作用。客观轨迹维度，用户信息需求由情感、认知、行动三大需求组成，并在对话中呈现一定时序递进；AI输出策略以影响反应为核心，情感支持与策略建议并重，认知重构占据重要比例；供需匹配程度是决定交互路径分化的核心枢纽，匹配时走向正向深化，错位时导致纠偏或中断。 基于上述发现，本研究构建了人智交互缓解心理压力机制模型，整合了压力缓解机制、主观体验层、交互过程层，揭示了从行为轨迹与主观意义到效果产出的完整机制，为理解压力状态下人智交互提供了系统的理论解释框架，为AI心理健康服务的优化提供了实践启示。
英文摘要：	With the rapid development of generative artificial intelligence and the growing prominence of mental health issues among college students, human-AI interaction has shown significant potential in the field of mental well-being. Existing research has primarily focused on users' acceptance and attitudes toward AI-based mental health services; however, the underlying mechanisms through which human-AI interaction alleviates psychological stress remain poorly understood. The alignment between user needs and AI offerings, subjective user experiences, and the mechanisms by which interaction reduces stress are still unclear. This study focuses on the interactive behaviors of college students under stress when engaging with generative AI. It employs a mixed-methods approach combining experimental methods, semi-structured in-depth interviews, and content analysis. We systematically analyze in-depth interview data from 25 participants and conduct content analysis on 25 sets of dialogues between users and DeepSeek, supplemented by quantitative assessments using pre- and post-stress measurements. The study unfolds progressively across three levels: effectiveness of stress reduction, construction of subjective experience, and objective behavioral trajectories. Findings indicate that human-AI interaction is effective in reducing stress. Stress relief occurs through six mechanisms: cognitive restructuring, emotional support, action empowerment, perceived support, decision clarification, and normalization—each pathway differing depending on the type of stress. At the level of subjective experience, user experience consists of five core components: information needs anticipation, information interaction behavior, perception of information presentation, evaluation of information utility, and assessment of information source credibility. These form a dual-path structure involving both information processing and relationship building. The information-processing path follows a supply-demand matching logic aimed at problem resolution, while the relationship-building path follows a trust accumulation logic leading to credibility assessment. These two paths operate in parallel and influence each other. Furthermore, users’ initial engagement with AI for stress relief is shaped by three forces: push, pull, and resistance. On the objective trajectory dimension, users’ information needs consist of emotional, cognitive, and behavioral components, unfolding in a sequential progression during dialogue. AI output strategies emphasize impact and response, balancing emotional support with practical suggestions, with cognitive restructuring playing a prominent role. The degree of supply-demand alignment serves as a critical pivot determining the divergence or convergence of interaction pathways—when aligned, interactions deepen positively; when misaligned, they lead to correction or interruption. Based on these findings, this study constructs a comprehensive model of the mechanism through which human-AI interaction alleviates psychological stress, integrating stress reduction mechanisms, subjective experience, and interaction processes. It reveals a complete mechanism chain from behavioral trajectories and subjective meanings to outcome effects, offering a systematic theoretical framework for understanding human-AI interaction under stress and providing practical insights for optimizing AI-powered mental health services.
查看全文：	预览  下载(下载需要进行登录)